The age discount: last season's model, this season's price
Model age is the quietest lever on an API bill. Across 108dated models, the median price falls from $2.83/1M tokens for models under six months old to $0.75/1M past two years — 73% less for capability that was state of the art when it shipped. Every figure is today's cheapest live offer.
Median price by age
Median blended $/1M tokens (cheapest live offer per model, $0 free tiers excluded), bucketed by time since release.
Every model, by age
One dot per model: its age today against its current price, log scale. The downward drift left-to-right is the discount accruing. Click a brand to isolate it; click a dot for its page.
Ages computed from sourced release dates, prices verified 2026-09-05. Survivorship caveat: old models still listed are the ones providers still find worth serving — retired models don't appear. Mirror view of the price trend by release date; see also how fast models get replaced · methodology.
FAQ
How much cheaper are older LLM models?
A lot. Across 108 models, the median blended price is $2.83/1M tokens for models under 6 months old, falling to $0.75/1M for models over two years — about 73% less. Age is the single biggest discount in the API market.
Is an older, cheaper model good enough?
Often, yes — for classification, extraction, summarization, and most high-volume production tasks, a model from last year performs indistinguishably from this month's release at a fraction of the price. The newest models earn their premium mainly on frontier reasoning tasks.
Why does age correlate with lower prices?
Three forces: providers cut prices on older models to keep them attractive, open-weight models accumulate competing hosts that bid prices down, and each new generation resets what a dollar is expected to buy. Survivorship also plays a part — old models that stay listed are the ones still worth serving.